InfEst is a Python library for estimating information measures from data.
From finite discrete data, InfEst can estimate:
- entropy
- joint entropy
- mutual information
- partial information decomposition (specific, unique, redundant, and synergistic information)
For the first three in the above list (and a simplified version of the partial information decomposition), InfEst uses the NSB method to provide less biased estimates for finite data. In these cases, uncertainties in the estimates are also provided.
To install the package and its python dependencies using pip, clone the repository, descend into the InfEst/ folder, and run
pip install -e .
For examples of code usage, see the included jupyter notebook file examples.ipynb.
Bryan C Daniels